Certain compensatory mutations at the protein level have been reported recently . Coordinated substitutions in NS3 and NS4A affect HCV replication via modulation of NS5A phosphorylation . Compensatory mutations in p7 and NS2 restore assembly-defective core protein mutants, whereas chimeric HCV with coordinated mutations in envelope 1, p7, NS2, and NS3 increase the intergenotypic compatibilities for virus assembly and release . More importantly, amino acid covariance networks have been identified to predict the response in HCV patients receiving anti-viral therapy . Such studies underscore the significance of the functional linkage of certain proteins and their covariant amino acid residues for HCV persistency, raising the possibility that molecular covariation can be computationally predicted during persistent infection for diagnosis, Foretinib biological activity prognosis and optimal drug selection. It is suspected that covariation might involve motifs in the UTRs which regulate HCV genome replication at transcriptional or translational levels and may be essential for persistent HCV. However, no studies have yet addressed covariation between the HCV UTRs and the NS proteins. In the present study, the authors explore the possibility that conserved covariation spots exist between functionally essential nucleotides in the UTRs and the amino acid residues in the 3 enzymatic NS proteins. The association data mining algorithm in the Weka software was used to extract previously unknown and potentially order PD 0332991 meaningful covariation within the HCV sequences retrieved from the Los Alamos HCV database at the full-length genome level . The functional relevance of the observed covariation sites was then tested in a cell-based HCV replicon system , analyzing the effects of either the individual or simultaneous substitutions of those sites with regard to replication efficiency and RNA-protein interactive ability. One of the most common applications of association rule mining is ��market basket�� analysis, i.e. a search is performed from supermarket checkout data for groups of items that occur together in transactions. A similar technique is used in this study, whereby the nucleotides and amino acid positions are considered as attributes in an individual instance. Association rule mining searches for covariation rules between single nucleotides of the UTRs and the amino acid residues of the NS proteins. To this end, 217 full-length HCV genome sequences were downloaded from the Los Alamos HCV sequence database on Nov. 30, 2006. Analysis of the phylogenetic relationships of the HCV sequences indicated that most were clustered into 4 major genotypes, 1a, 1b, 2a and 2b, while the others sporadically presented as 14 minor genotypes .